THE STEREOCHEMISTRY OF SQUARE COMPLEXES' DAVID P. MELLOR Department of Chemistry, University of S y d n e y , N e w South Wales Received February 8,1949 Traditional chemical methods of unravelling questions of molecular structure fail to provide a unique solution to the problem of the structure of bivalent platinum compounds. The various alternative configurations which may be used to explain the phenomena of geometrical and mirror-image isomerism among platinous complexes can be successively narrowed down by different chemical and physical methods of investigation until but one possibility remains-namely, square coordination. The theory of square coijrdination makes possible the correlation and interpretation of such a large mass of data on the compounds of bivalent platinum and certain other metals that there can be no reasonable doubt of the reality of this type of structure. With the reservation that i t applies only to bonds free to arrange, the theory of the directed valence bond makes i t clear just what metal atoms are likely to form square bonds.
I. INTRODUCTION In the design of molecules Kature uses relatively few fundamental units of pattern or configurations, and it is naturally important to know what these are and under what circumstances any particular configuration is likely to be found. We shall here be concerned with the narrow field of stereochemistry which deals with the orientation in space of the four valence bonds of quadricovalent elements. The tetrahedral disposition of four valence bonds about the carbon atom was the first configuration to be discovered, and so widespread did its occurrence eventually prove to be, not only among the compounds of carbon, where it was universal, but also among some forty other elements, that attention to this type of structure long overshadomed that given to an alternative arrangement of four bonds in space-namely, the square arrangement. Indeed, the tetrahedral arrangement proved to be so dominant a feature of molecular architecture that the very existence of the square pattern was,-and, for that matter, in some quarters still is,-a matter of controversy. Kevertheless the evidence on this question has grown steadily over the last few years and the time is opportune for some attempt to evaluate the significance of the square configuration for the stereochemistry of the metals. I n attempting to do this, two main questions will be considered: firstly, how strong is the evidence for the square disposition of valence bonds, and secondly, with what elements and under what circumstances does it occur? In tracing the developments leading to the proof of square coordination, attention will be centered on the compounds of platinous platinum (Pt") , because it is with these that the most abundant and satisfactory evidence has been obtained. Notwithstanding this, there has been a good deal of controversy about the structure This article is based largely on the Presidential Address t o the Royal Society of New South Kales, April 1, 1942. 137
138
DAVID P. MELLOR
of platinum compounds. Some idea of the controversial nature of this field may be gathered from a few excerpts, chosen a t random, from the recent literature: “Platinum salts combine with an enormous number of inorganic and organic groups or molecules and many of these salts have been known for a hundred years, yet the constitution of the isomeric diammines of the type [PtA2X2]is still a subject of controversy” (17). “It is still an open question whether the experimental work (relating to planar and tetrahedral structures) has been correctly interpreted or not or whether some elements can assume more than one structure” ( 5 ) . “In any event, the problem presented by these salts (certain allegedly optically active platinous salts) does not seem to be finally cleared up” (135).
Although, in a certain sense, no scientific problem is ever cleared up, there is reason to believe that the main outlines of the stereochemistry of bivalent platinum are well established and it is hoped to show that the issues raised in these quotations have now been largely settled.
11. THE DISCOVERYOF THE SQUARECONFIGURATION It is not surprising that the square configuration was first discovered among the compounds of Pt”, because, as we now know, no other element forms so many compounds which exhibit isomerism on account of this stereochemical characteristic. The two substances primarily concerned with the development of our knowledge in this field,-namely, the a- and p-forms of dichlorodiammineplatinum, [Pt(NH3)2C12],-were discovered nearly a century ago, the former by Peyrone (123) and the latter by Reiset (133).2 The methods used in their preparation involve two similar processes; the a-compound is made by replacing two chlorine atoms of the [PtC14]-- ion with ammonia molecules, the @-formby replacing two ammonia molecules of the [Pt(NH3)4]++ion with chlorine atoms. The latter operation is effected either by heating solid [Pt(NH3)4]C12under carefully controlled conditions or by treating the aqueous solution with concentrated hydrochloric acid; the former, by treating K2PtC14 with aqueous ammonia. There is little doubt that each of the above substitutions proceeds stepwise: Tu” YH [PtCl4]-- -Bt [Pt(I%”3)Cls]- X [Pt(KH3)2Clz]O cy
c1-
[Pt(NHs)4]++ ---+
c1-
[Pt(NHa)aCl]+ -+=
[Pt(NH3)2CLlo
P
The intermediate compounds have been isolated and each step has been carried out separately. 2 Since the discovery of these compounds, considerable confusion has arisen in regard to their names. They were first known as the chlorides of Peyrone and of Reiset, respectively, and later as plato semi-diammine and platosammine chloride. In 1893 Werner introduced the terms a and 8 ; finally, Drew and his collaborators, for no very good reason, reversed the usage of a and p. I n this article Werner’s nomenclature will be retained.
STEREOCHEMISTRY O F SQUARE COMPLEXES
139
The problem of explaining the existence of the a- and @-compoundsresolves itself, as a first step, into deciding whether they are (a) isomers (structural or geometrical), or (b) polymers, or (c) dimorphs. These alternatives, though not explicitly formulated by the earlier workers, can, as a result of their work, be narrowed down. The last was eliminated first. Cleve (22), a very active early worker in this field, clearly established the different chemical behavior of the aand p-forms of [Pt(NH3)&12]. By treating each form with a series of reagents, including the appropriate silver salts, he prepared and described new (isomeric) compounds such as the a-and @-formsof [Pt(NH&Brz], [Pt(NH3)212], [Pt(NH& (CN)z], [ P ~ ( N H & ( N O ~ )and Z I , [Pt(NK3)2(NOz)zl. Although this and later work leaves no doubt that the a- and p-forms are not just simply different crystalline modifications of the same substance, it is interesting to note recent confirmation along physical lines. Dimorphous molecular crystals contain the same molecules packed in different ways, so that when each crystal structure is broken down by solution in any given solvent, the resulting solutions should be identical, a point which can, for example, be tested by an examination of absorption spectra (105). Small, but definite, differences have been noted in the absorption spectra of the aqueous solutions of the a- and @-formsof [ P ~ ( N H ~ ) z C (4). I Z ] There is little doubt that, had chemical and physical tests along these lines been carefully applied, many of the issues created by the announcement of an alleged third (r)form of [Pt(NH3)&12](43) would have been avoided. If we make the assumption that the coordination number of Pt" is four,' and experience ha's shown this to be practically universal, the number of polymers of the empirical composition [Pt(NH3)zClz]z is, for 2 > 1, limited to the following:
1. [Pt(NH3)4][PtCL] (88) 2. [Pt(NH3)&1][Pt(NH3)C13] (122, 124) 3. [P~(T\TH~)~][P~(NHI)C~~]Z (23) 4. [Pt(NH3)&1]2 [PtCL] (22) The a- and @-diamminesare distinct from all of these. Anticipating the discussion in a subsequent section, one may add that, with one rather puzzling exception, molecular-weight measurements on a- and 0-forms of compounds pyridine, C2H6NH2,etc., and X = of the type [PtAzXz] (where A = "3, C1, Br, CNS, OH, etc.) shorn that both forms are monomeric. 3
Unless this limitation is specified i t would be necessary to consider a structure like: C1
....
NH3C1
\I/
/qt\ C1
NHO C1 NHj
\I/
,?t,
NH3C1
\I
/9t
....
TiHaC1 NH3
There is, however, no evidence for this structure among any platinous compounds of the empirical composition Pt(NHs)&12. Some extremely rare instances were Pt" may possibly be octahedrally coijrdinated will be referred to later.
140
DAVID P. MELLOR
With the elimination of the possibilities ( b ) and (c)-polymerism and dimorphism, respectively-the problem now resolves itself into deciding whether the a- and p-diammines are structural or geometrical isomers. One of the first attempts to understand their constitution was made by Cleve (22), who proposed structures which, in the light of the then prevailing theories of valency, seemed plausible enough. They were : NHs-NH3 -C 1
NHa-Cl
/ Pt \
/ Pt \
NH3-Cl
c1
a
P
These formulations, which were also supported by Blomstrand, Jorgensen, and others, implied, of course, that the substances were structural isomers. With present-day knowledge of atomic structure, which enables an upper limit to be placed on the number of covalent bonds that can be formed by first-row elements of the Periodic Table, these structures can be ruled out immediately, since they both involve five covalent bonds to nitrogen. As will appear in the sequel, there are many other reasons for rejecting them. Xevertheless some attempt was made to revive them a few years ago (43) and they are occasionally still seriously discussed in the literature. The revival was the result of an attempt to explain certain reactions of the two compounds but, without going into detail, it can be stated that all these reactions can equally well be explained on an alternative view of their constitution. While Cleve's formulations of the two compounds are no longer tenable, they served a very useful purpose in focusing attention on the problem of their constitution. Some fifteen years after Cleve's work, Jorgensen (77) set out to determine experimentally whether the assignment of the structures to the a- and p-forms as above had been correct, and in so doing he laid the experimental foundation upon which one of the most important advances in our knowledge of the structure of platinous compounds was made. This was Werner's introduction of the hypothesis of square coordination. Rejecting all previous explanations in terms of structural isomerism, Werner (152) applied to the problem the principles which he so successfully used to account for the constitution of the cobaltic ammines. Realizing that, unlike cobaltic ammines, which were universally characterized by a coordination number of six, the compounds of platinous platinum were always four coordinated, he put forward the idea that the a- and p-diammines were geometrical isomers (cis and trans), owing their existence to a planar distribution of the four bonds about the platinum atom, as shown below : 3"
\ / Pt / \
"3
a (cis)
C1
NH3
c1
c1
\ / Pt / \
c1 NH3
p(trans)
141
STEREOCHEMISTRY O F SQUARE COMPLEXES
With regular tetrahedral bonds from platinum, two isomeric diammines are not possible. Although it would seem that Werner never a t any time explicitly stated that the coplanar bonds were directed towards the corners of a square, it is clear from the diagrams of his classical 1893 paper that he considered them to be so directed. He never stressed the size of the bond angles in the plane, presumably because it was not essential in explaining the geometrical isomerism. Actually, the term “spume coordination” was first used by Pauling (117) in connection with his quantum-mechanical treatment of the directed valence bond. Strictly speaking, any one of a number of structures would equally well account for the geometrical isomerism of the a- and 8-dichlorodiammines. The four bonds from platinum might be directed towards: (a) the corners of a tetragonal or rhombic bisphenoid (figure la shows the former); (b) and ( c ) the corners of a square (figure lb) or a rectangle (figure I C ) ; (d) the corners of a tetragonal or rhombic pyramid (figure I d shows the former). Each one of these alternatives
(a)
(b)
(4
(4
FIG.1. Possible structures to account for the geometrical isomerism of the a-and@-forrns of dichlorodiammineplatinum.
has been introduced from time to time to explain the results of some chemical investigation. Thus, the structure shown in figure l a was discussed by Rosenheim and Gerb (38) in explaining the existence of certain supposedly optically active platinous and palladous complexes. Drew and his collaborators (44) appear to have had the configurations shown in figures l a and I C in mind when formulating the hypothesis of paired coordination links to account for the structure of certain tetrammines. The pyramidal structure, figure Id, was suggested by Dwyer and Mellor (45)as a means of reconciling results of certain experiments on mirror-image and geometrical isomerism. It does not seem to have been widely realized that the results of purely chemical methods of investigation (study of composition, isomerism, reactions, etc.) fail to provide a unique solution to the problem of the structure of platinous compounds and that herein lies the origin of much of the controversy in this field. To distinguish between the alternative structures (figure 1, a to d) it is necessary to have some information about the sizes of the angles between the four platinum bonds. Since the phenomena of geometrical and mirror-image isomerism depend primarily on molecular symmetry, they reveal nothing about the dimensions of bonds or the sizes of the angles between them. For optical activity to appear in the molecule CRIR2RsR4, all that is necessary is that the bonds have a general tetrahedral orientation; there is no need for the angle of the bond
142
DAVID P. MELLOR
R1-C-% to be 109"28',-it might be 140" or more. Similarly, the existence of cis and trans isomerism among platinous compounds cannot be used to draw any inferences about the size of bond angles; any one of the above models might equally well be used to account for the geometrical isomerism. In view of the impasse which confronts the chemical method, it would perhaps be more logical to pass on to physical methods of studying the problem. This course will not be followed, partly for historical reasons and partly because, although the chemical evidence does not permit a single unequivocal choice, it does allow the alternatives to be narrowed down. This last can be achieved by combining the evidence from the study of geometrical and of mirror-image isomerism. To avoid unnecessary repetition of qualifying statements, the correctness of Werner's hypothesis will be assumed in the following sections dealing with geometrical isomerism.
111. GEOMETRICAL ISOMERISM A. T H E DETERMINATION OF T H E CONFIGURATION O F T H E DIAMMINES
Having decided that the a-and p-diammines were geometrical isomers, Werner went a step further and by very ingenious reasoning determined which of the two forms was cis and which trans. The reactions concerned, first studied in detail by Jorgensen (77), may be summarized under two headings: 1 . Addition reactions
When treated with two molecules of pyridine, a-dichlorodiammineplatinum forms a-diamminedipyridineplatinous chloride. This latter compound can also be prepared by treating a-dichlorodipyridineplatinum with two molecules of ammonia : [Pt(KHB)zClzl ~ P Y
+
a
\
I
//*
[Pt("3)
z(
a
P YIC12 ~
(1)
[Pt(PY)zClz] -k 23" a
Similar reactions are observed with p-dichlorodiammineplatinum:
2. Elimination reactions4
When the solids are heated alone or aqueous solutions of them are warmed with concentrated hydrochloric acid, the a- and p-tetrammines revert to the 4 These reactions might also be called substitution reactions if we regard them from the point of view of the complexion only. The names have been used in reference t o the molecules as a whole.
143
STEREOCHEMISTRY OF SQUARE COMPLEXES
dichlorodiammines. Thus a-dipyridinediammineplatinous heated, reverts to P-dichloropyridinemonammineplatinum : [Pt(”s)z(~~)zIClz
+
+
[Pt(~H3)(py)Clzl
P
a!
“3
chloride,
+ PY
when
(31
On the other hand, 0-dipyridinediammineplatinous chloride gives rise to a mixture of P-dichlorodiammineplatinum and p-dichlorodipyridineplatinum :
-2PY/ [Pt(”3)
z (PY)z IClz
P
7
[Pt(”3
P
1-2C
~1 P
/ \
- 2NHa\
(4)
L
[Pt(PYh Clzl l3
The last reaction has been queried by Reihlen and Nestle (131), but the experimental work of Jorgensen (77) and of Drew et al. (44) shows that the reaction which takes place is the one formulated. The products were separated by fractional crystallization and identified by preparing distinctive derivatives. All other addition and elimination reactions have been checked by Drew et al., who used them, not to confirm Werner’s hypothesis, but as a basis for the hypothesis of paired coordination links. Suffice it to say that the experimental foundation upon which Werner built his “configuration determination” has stood the test of time. Whereas Jorgensen was forced to introduce several arbitrary assumptions to account for the above reactions in terms of structural isomerism, Werner was able to account for them in a perfectly straightforward manner with the help of only one further assumption, oiz., “trans elimination.” Werner pictured reactions 1 to 4 as proceeding in the following manner:
a (cis)
Clz a (cis)
a (cis)
144
DAVID P. MELLOR
\ / Pt
-
/ \
" 3 c1 @(trans)
- H3N
\
/py
Pt
ICl2
@(trans) In attributing the cis structure to the a-compound and the trans structure to the P-compound as above, Werner's final conclusions have been anticipated. His argument in support of this assignment is based on the assumption that, in the course of the elimination reactions, pairs of groups in trans positions only, are removed. The results of trans elimination are made clear in the following diagrams where the dotted lines enclose the eliminated trans pairs:
/i?(frans) An examination of the structure of cr (cis)-dipyridinediammineplatinous chloride will show that elimination of pairs+f cis groups should result in the formation of a mixture of three compounds, viz., [Pt(NH3)2C1z],[Pt(py)2C12], and [Pt(py)(NHI)C12],whereas actually only one, the last, is obtained. On the other hand, cis elimination from the @ (trans)-dipyridinediammine complex should result in the formation of only one compound, viz., [Pt(py)(KH3)C12]; actually two are found, [Pt(py)2Clz],and [Pt(NHs)2C12]. If we accept the hypothesis
STEREOCHEMISTRY OF SQUARE COMPLEXES
145
of trans elimination, all the reactions find consistent interpretation in terms of a cis structure for a-[Pt(NH&C12] and a trans structure for p-[Pt(NH3)2Cl~]. Such then is the experimental foundation upon which Werner, duly acknowledging his debt to Jorgensen, built the planar hypothesis. Yet nearly forty years later, protagonists of the view that the diammines were structural isomers charged Werner with “ignoring the relevant chemical evidence of his predecessors” when dealing with this problem (2). Perhaps the only weakness in the interpretation of the elimination reactions is that a t one stage (elimination from the a(~is)-[Pt(py)~(NH~)&l~)] it depends on a negative result,-failure to find more than one compound. It is a striking tribute to Werner’s remarkable insight into the structure of coordination compounds that all subsequent determinations of the configuration of a- and p-dichlorodiammineplatinum have proved the correctness of his assignment of cis and trans structures, and a t the same time justified his hypothesis of trans elimination -at least for these reactions. B. FURTHER CHEMICAL EVIDEKCE CONFIRMING WERNER’S ALLOCATION OF
CONFIGURATIONS
With the exception of results from dipole-moment studies and from one or two incomplete x-ray crystal analyses, confirmation of Werner’s work on configuration has been obtained along chemical lines. I n the course of extensive investigations of “ammoniacal platinum bases,” Cleve (22) reported a very interesting difference in the behavior of the a- and p-forms of [Pt(NH3)2(X03)2] towards oxalic acid solution: the a-form was converted to a compound with the empirical composition [Pt(NH3)2Cz04]; the p-form to a compound with the composition [ P ~ ( N H ~ ) Z ( C ~ O Some ~ H ) ~sixty ]. years later Griinberg (60) confirmed these observations and first suggested their interpretation. In doing so he made use of a method developed by Werner in his study of cis-trans configurations among the octahedral complexes of CoI’I. Thus, Griinberg proceeded on the assumption that the ( 3 2 0 4 - group acted-as a bidentate chelate only when it replaced two NO3 groups in cis positions; when it replaced NO3 groups in trans positions, the oxalic acid molecule occupied one coordination position. It is clearly sterically impossible for the C204-- group to span trans positions if square bonding is to be maintained.5 Since the oxalate group enters a-[Pt(NH3)2(N03)2]as a bidentate chelate, this form obviously has the cis configuration. X o w a-[Pt(NH&(XO&] is prepared by treating the corresponding (a) chloro compound with silver nitrate, and if it be assumed that the substitution of NO3 for C1 occurs without change of configuration, then a-[Pt(NH3)2C12] must also be a cis-form, in agreement with Werner’s contention. That no change of configuration does occur is shown by the behavior of the oxalato compounds towards hydrochloric acid: the oxalato compound made from cis-[Pt (NH3)2(N03)2] regenerates cis-[Pt (NH&C&], while the second oxalato compound regenerates The existence of the so-called trans-Pt(py)~S04would seem to be an exception t o this statement. The compound is, however, a dihydrate and is probably [Pt(py)2(HZ0)2]S04.
146
DAVID P. MELLOR
the trans-dichloro compound. These reactions may be represented schematically:
[
c1
"s NH C 3']
I
3"
c
\ / Pt / \
HCl
LHOOCCOO
3"
oCocooH "8
1
A very similar cycle of reactions has been carried out with a(cis)-[Pt(py)zClz] (43). [Pt(PY>2(OW21
\
7
\H2C204
L
Cis(a)-[Pt(py)2Clz] also reacts with dipyridyl (dipy) to form [Pt(py)2dipy]Cl2 (114), whereas prolonged boiling of the trans(p)-form with alcoholic dipyridyl solution produces no detectable reaction (98). Passing reference only can be made to other work on the configuration of the dichlorodiammines; further confirmatory evidence is to be found in Griinberg and Ptizyn's (61) work on glycine complexes 3"
/ \
PtC14 HOOCCH2 NHs 2
\ / Pt / \
NH2 CH2 C 0 OH
3"
Clz
147
STEREOCHEMISTRY OF SQUARE COMPLEXES
and also in some work by Drew (41), who has shown that a(cis)-[Pt(NH3)2C12] reacts with ethylenediamine to form [Pt(NH3)2(NH2CH2CH2NH2)]Cl2,
whereas no reaction occurs with the trans-form. Another method of discriminating between cis- and trans-[PtAeX2] complexes depends on Kurnakow's (82) test, which is based on the reactions: th thiourea (th)+
c1 .(cis)
These reactions, in themselves, tell one nothing about structure in the sense that the ring-closure reactions do. Kurnakow's test has been very frequently used by Russian workers for deciding on the structure of platinous ammines. It would be interesting to know how widely applicable the test is, and it would be worth while to check it against other substances of known configuration, such as the thioether, stibine, arsine, and phosphine derivatives, described by Jensen (71, 73). C. RELIABILITY OF THE CHEMICAL METHOD OF DETERMINING CONFIGURATION
A check on the reliability of the chemical method is afforded by dipole-moment measurements, which will be dealt with more fully in a subsequent section. Suffice it to say here that confirmatory evidence from this source has been obtained with t r ~ n s - [ P t ( p y ) ~ Cand l ~ ] with both forms of [Pt((CzH&S)2CL]. The reactions of C ~ - [ P ~ ( ( C ~ H ~reported ) ~ S ) ~ Cby~ Angell ~ ] ~ et al. (2) show that it has a cis structure; the measured dipole moment of the compound bears this out. Dipole-moment measurements could not, unfortunately, be extended to the ammonia complexes, owing to their limited solubility in non-polar solvents. One interesting point emerges from all this work and that is the remarkable stability of the configuration of platinous ammines, a circumstance which has greatly facilitated their study. There is no evidence that rearrangement of atoms or groups relative to one another occurs in a dichlorodiammine during 0
This compound is referred t o as the @-formby Angell and his collaborators.
148
DAVID P. MELLOR
chemical reaction, that is, a trans-compound retains its trans configuration through a series of reactions.' As Wells (150) has shown in his study of the structure of A ~ [ C O ( N H ~ ) ~ ( N this O ~ )constancy ~], of configuration is not characteristic of all complexes. It is certainly not characteristic of square palladous complexes which, for some reason as yet unknown, are not as robust, in this sense, as those of platinous platinum. D. MOLECULAR-WEIGHT DETERMINATIONS ON
PtAzXz COMPLEXES
Although all the possible polymeric forms of compounds with the empirical composition Pt (NH3)2C12have already been discussed, it is necessary to refer to this topic once again, because one of the first criticisms of the theory of square coordination originated in some work on molecular-weight determinations.
TABLE 1 Molecular weights and melting points of platinous compounds of the type [PtAeX2] Jensen (7 ~~
YOLECULAP WEIGHT
SUBSTANCE
c
MELTING POINT
Found
Calculated
"C.
trans- [PtC12(Et2Se) 21. .......................... cis-IPt C12 (Et~Se)l].............................
60.0 74.0
549.0 618.0
540.7 540.7
{cis-[PtCI2(Pr2S) ..............................
........................... trans-[PtCI2(Pr~S)2].
80.0 85.0
531.0 598.0
502.5 502.5
trans-[PtCI2(Bu2S)2] ........................... ci~-[PtClz(B~tS)21 .............................
60.0 84.5
564.0 612.0
558.4 558.4
110.0 138.5
538.0 672.0
558.4 558.4
21
1trans- (PtC12 (i-Buns) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \cis-[PtC12(i-Bu~S)~] .............................. , Et = ethyl; P r
=
propyl; Bu = butyl; i-Bu = isobutyl.
From a careful study of the vapor-pressure lowering in liquid ammonia solutions, C l ~dimeric ] in that Reihlen and Nestle (131) showed that t r a n ~ - [ P t ( N H ~ ) ~was solvent. At the same time they showed that both forms (cis and trans) could be recovered unchanged from liquid ammonia and proved their technique by measurements on known substances. As the authors themselves point out, the dimerism of the trans-form does not prove the square hypothesis false. In spite of this and the fact that Fritzmann (53) had shown that both forms of [Pt{ Se(CsH11)2)zClz] were monomeric, Reihlen was SO convinced that Werner's interpretation of the structures of the diammines was incorrect that he began a long series of experiments on optical activity, and eventually became one of the chief opponents of the theory of square coordination. 7 This is, of course, not true of all platinous complexes; i t has been reported, for example, that interconversion of cis- and trans-forms of various thioether complexes occurs quite readily (2).
STEREOCHEMISTRY O F SQUARE COMPLEXES
149
An ebullioscopic determination of the molecular weight of trans-[Pt (NH3)2C12] in aqueous solution, admittedly not very accurate (43), shows it to be monomeric in this solvent. King (80) has shown that the molecular weights of both forms of [Pt(NH3)2(0H)z] are normal in aqueous solution. As far as other compounds of the type [PtA2X2]are concerned, there is plenty of evidence that both isomeric forms possess normal molecular weights. One may cite the work of Hantzsch (63) on CY- and @-[Pt(py)ZClz],of Grunberg (59) on a-and /3-[Pt(NH3)2(CNS)Z], and finally the very extensive of Angel1 et al. (2) on CY- and @-[Pt((C2H6)zS)zC12], series of measurements by Jensen (71) on a wide range of thioether complexes (table 1). All this work leaves no doubt that the cis- and trans-compounds are monomeric in the solvents used. I n Jensen's work there is some evidence of association of the cis-forms, but only to a small degree and in keeping with what one would expect from their higher dipole moment. It is interesting to note, too, the effect of the dipole moments on melting points; of the two, the cis-form melts a t the higher temperature. The problem of the dimerism of trans-[Pt(NH&C12] in liquid ammonia still remains. If dipole moment were the principal factor determining association, one would expect the cis-form to be the one more associated. Any mechanism of association involving weak hydrogen bonds should operate with both isomers and might reasonably be expected to cause association beyond the dimer stage:
. .Cl
\
...HaX
H N-H /H ~~
I*
Pt
/ \
........c1
C1........H-x
H
..
"3.
\ / Pt H/ \
Cl..
.
All that can be said at present is that the association of the trans-form in liquid ammonia is peculiar to this isomer and to this particular solvent, and, while inexplicable a t the moment, it does not weaken the theory of square coordination. The importance of the work of Reihlen and Nestle lies not so much in the peculiarity of the results obtained as in the stimulus it gave to work on the optical activity of platinous compounds, to which we shall now turn.
IV. THEMIRROR-IMAGE ISOMERISM OF PLATINOUS COMPOUNDS If two unsymmetrical chelate groups such as isobutylenediamine are tetrahedrally coordinated to a central metal atom, the resulting complex may exist in mirror image (figure 2a) but not cis-trans forms. On the other hand, if the chelate groups are coplanar, the complex may exist in cis-trans, but not mirrorimage forms (figure 2b). Finally, if Ptrl forms pyramidal bonds, both geometrical (cis-trans) and mirror-image isomerism are possible ; of the geometrical isomers, the trans-form only is capable of existing in mirror-image isomers. Over a number of years, Reihlen and his collaborators, working with various unsymmetrical chelate molecules (see figure 2a), have made repeated claims that they have obtained evidence proving the tetrahedral configuration of Pt".
150
DAVID P. MELLOR
owing to he fact that these claims clash with the main body of evidence, it is necessary to examine them in some detail. In the first place it must be pointed out that, except in one very doubtful instance, Reihlen and his collaborators have never followed the classical method of establishing a configuration by resolution through to its proper conclusion. That is to say, the allegedly optically active complexes have never been obtained free from the resolving acid (or base). In the one instance (132) where this
A -... . . -.. ..-,. I
.
*...
(Cl FIG.2. Possibilities of isomerism when two unsymmetrical chelate groups are coijrdinated to a cent,ral metal atom. The arrow in these diagrams is used to indicate a molecule which by chelation produces an unsymmetrical ring, e.g.,
NH2-CHz
I
NHz-C(CHd2
separation was reported as having been carried out, the observed rotation was so very small (0.06’ for a 4-dm. tube) as to be without any special significance. Reihlen’s claims, then, rest on the observation that the molecular rotation, [MI, of a salt like bis(aminomethyl-3-ethyl-4-methylquinoline)platinousa-bromocamphor-n-sulfonate (130) is greater (or less) than that calculated from the a-bromocamphor-7-sulfonate content of the salt. All attempts to obtain an optically active complex free from the acid used for resolution resulted in failure, which was attributed to racemization. In view of the stability of platinous
STEREOCHEMISTRY O F SQUARE COMPLEXES
151
complexes, already noted in an earlier section, this explanation is not specially convincing. Unsuccessful attempts to obtain optically active platinous complexes have been reported by other workers. Ostromisslensky and Bergman (116) failed with [PtClNH3SOs(py)]-, Tscherniaev (144) with [Pt(NH20H)(NH3)(py)(NOz)]+,and finally Jensen (76) with bis(thiosemicarbazide)- and bis(2-aminomethyl-3-ethyl-4-methylquinoline)platinousions. Significantly enough, in the last instance Jensen did note rotation differences of the same order as were observed by Reihlen and Huhn (130). However, on attempting to isolate an optically active platinous complex by precipitation with picric acid, Jensen found that the chloroform solution of the resulting picrate mas always entirely devoid of optical activity. It is worth noting that several of the compounds upon which attempts at resolution were made have been described in the isomeric forms to be expected from square bonding: thus, two forms of bis(isobuty1enediamine)platinous chloroplatinite (42) and bis(phenylethylenediamine-2-aminomethyl-3ethyl-4-methylquinoline)platinouschloride (132) and three forms of [Pt(NH20H)("3) (py) (NO,)]+ (143, 144) have been isolated. In the belief that Reihlen's resolutions were valid, Dwyer and Mellor (45) suggested a pyramidal structure for the complexes of platinum and palladium with a view to reconciling the findings on optical activity with the known cis-trans isomerism. This view is, of course, no longer tenable. There still remains the question as to what causes the changes in rotation discovered by Reihlen and Jensen. The latter has made the suggestion that the bromocamphorsulfonate molecule becomes attached to the platinum atom itself, either ( 1 ) by forming the complex I
I or (2)by opening the chelate rings to form 11: Ci 2 Hi 2 NCHz NH2
\ /
S OSC1o H14 OBr
Pt
Jensen has calculated that contamination of bis(2-aminomethyl-3-ethyl-4methylquino1ine)platinous a-bromocamphor-n-sulfonate with as much as 5 per cent of I makes so very little difference in the analytical figures that it would not be detected. Formation of substance I1 would not alter the analytical results. Experiment shows (76) that on the addition of hydrochloric acid to
152
DAVID P. MELLOR
bis(2-aminomethyl-3-ethyl-4-methylquinoline)platinous a-bromocamphor-rsulfonate (which may be contaminated with I or 11, or both), the molecular rotation returns to its normal value, i.e., the value for the free bromocamphorsulfonic acid. Jensen explains this behavior as being due to the freeing of the bromocamphorsulfonic acid from I and/or 11. Whether we accept Jensen's explanation of the rotation changes or not, the fact remains that none of the foregoing optical evidence can be said to favor the view that Pt" is tetrahedrally coordinated. Rather does it constitute negative evidence for alternative structures, but such evidence is seldom very convincing because we cannot be certain that the cause or causes of the negative result are exactly what we imagine them to be. The unsatisfactory situation of the resolution work just described has been largely cleared up by Mills and Quibell (111),who were the first to describe stable optically active derivatives of Pt". Unfortunately, as far as the unequivocal proof of the square structure goes, the work of Mills and Quibell on mirror-image isomerism leaves us in much the same position as does the work on geometrical isomerism; that is to say, their results are quite consistent with the square structure, but they may be explained in terms of other configurations. As they were doubtful of the interpretation of the work' on geometrical isomerism, Mills and Quibell planned and achieved the synthesis of a very ingeniously devised complex which, if planar, would have the symmetry properties required to produce mirror-image isomers. As represented below "2
"2
/
+---Ph4
\
CH
CH2
I-... +
Pt
C-Me
/ Ph
\ "2
"2
Me
the molecule possesses neither a plane nor a center of symmetry and is therefore resolvable. On the other hand, if the chelate group on the left (meso-stilbenediamine) were fixed, while the chelate group on the right were rotated through 90') so bringing about a tetrahedral distribution of the four bonds to platinum, the molecule as a whole would possess a plane of symmetry. In other words, the molecule with a tetrahedral configuration would no longer be resolvable. I n point of fact, Mills and Quibell were able to isolate, quite free from the resolving acid (d-diacetyltartaric), optically active salts whose great stability is entirely in keeping with what is known of the robustness of platinous complexes. And so for the first time the method of resolution was used to provide evidence for the square configuration. I n the light of the general body of evidence, more especially the physical, there is no doubt that these optically active complexes 8 "Although the frequency with Which isomerism occurs in compounds containing a complex of the type AZPtB2 gives great weight to Werner's interpretation, yet it is always difficult in dealing with geometrical isomerism to make certain that the isomerism is actually of the nature supposed. There is no such difficulty with mirror image isomerism" (111).
153
STEREOCHEMISTRY O F SQUARE COMPLEXES
are square, but as Mills and Quibell themselves point out, the results might also have been attributed to a pyramidal arrangement. However, they regarded this last configuration as inherently improbable, pointing out that if it were the explanation of their results, then certain simpler complexes should be resolvable. Inherent improbability is not necessarily a safe argument, as the recent discovery of the pyramidal arrangement of four covalent bonds about Pb" (113) will serve to show. Summing up the chemical evidence, one may say that it definitely rules out the regular tetrahedral and tetragonal bisphenoidal arrangements, is consistent with the theory of square coordination, but leaves the possibility of pyramidal and rectangular arrangements in doubt. For the final stages of the proof of structure we must turn to physical methods.
v. RAMANAND
INFRARED
SPECTR-4
In applying Raman and infrared spectra to unravelling problems of molecular structure the procedure is, briefly, as follows : Various molecular models are set up and the modes of vibration of each are systematically examined. According TABLE 2 R a m a n and infrared spectra of molecules of the type A(XY)4 YODEL
Square. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tetrahedral. . . . . . . . . . . . . . . . . . . . . . . . . . . . Pyramidal (square base). . . . . . . . . . . . . . . . Rectangular. . . . . . . . . . . . . . . . . . . . . . . . . . . . Pyramidal (rectangular base) . . . . . . . . . .
SYMMETRY
INFRARED FREQUENCIES
1 ~
IN.44YE
I
COINCl-
QUENCIES
6
4 9
1 1
12 16
0
2 ~
1
j
0 16
* The figures in parentheses refer t o the number of polarized lines. as the modes of vibration produce changes in the degree of polarizability or dipole moment of the molecule, the corresponding vibration frequencies will be responsible for lines in the Raman and infrared spectrum, respectively. The number of lines in each spectrum and the state of polarization of the Raman lines are deduced for each model and the model adopted is the one which gives exact or closest correspondence with the observed spectra. The number of Raman lines alone is not always a certain guide in deciding between alternative structures because there is always the risk that a line, though permitted, may be so faint as to escape detection. In the general case, if reliance is to be placed on structural conclusions, observations must be made on the polarization of the Raman lines and also on the absorption spectrum in the infrared region. As far as square complexes are concerned, the simplest spectra are to be expected from an ion like [PtCL]-- but, owing to difficulties associated with observing the Raman spectra of highly colored solutions, K2[PtC14] has not been investigated. Mathieu (96), who has been mainly responsible for work in this field, has studied the following colorless or nearly colorless compounds :
154
DAVID P. MELLOR
NadPt (CN141, Naz[Pt (NOd41, [Pt(NHd41C12, [Pt(py)4lC1~, and [Pt(en)~IClz; of these, only the first provided sufficient information for any structural conclusions to be drawn. Observations were made on the number and polarization of the Raman lines and were discussed primarily from the standpoint of square and tetrahedral models. The observed Raman spectrum of Naz[Pt(CN)41 comprised six lines, two of which were polarized; these data favor the square structure but do not constitute decisive evidence for it. The position can best be understood by reference to table 2', which shows the frequencies to be expected for different models of a molecule of the type A(XY)4 with the atoms AXY collinear. It will be seen a t once that the experimental data rule out the last three less symmetrical configurations. A clear-cut distinction between the first two models, in favor of the square structure, is most likely to be obtained by extending observations to the infrared, where the observation of five or more fundamental frequencies would definitely exclude the tetrahedral model. As the position stands, the evidence for the square structure is essentially negative in character; the number of Raman lines does not exceed that predicted for the square model. It is of course always more satisfying to have positive evidence on a point like this and there is a good case for extending work to the infrared. VI. EVIDENCE FROM DIPOLE MOMENTS Whereas all diatomic molecules of the type A-A are electrically symmetrical and non-polar, those of the type A-B always have a permanent electric moment. I n a polyatomic molecule, each bond A-B is associated with a dipole moment which can be treated as a vector quantity, the permanent dipole moment of the molecule as a whole being the vector sum of the individual bond moments. For any molecule with a center of symmetry, the vector sum of the bond dipole moments will be zero, that is, the molecule will have no permanent dipole moment. This is found to be true of a number of molecules whose structures have been ascertained by other means; thus CO,, SFe,and trans-dichloroethylene are all centrosymmetrical and without permanent dipole moment. In using dipole moments as a test for the presence of a center of symmetry it must be remembered that the converse of the above proposition is not necessarily true, that is to say, there are molecules like CC14, with zero dipole moment, which do not have a center of symmetry. However, of the configurations listed on page 153 only two have a center of symmetry; for all the others the vector sum of the dipoles will not, in general, be zero. Among the dipole studies of metal complexes, those of Jensen (71,73), dealing with isomerides of the type [PtA2X2],have an important bearing on the present discussion. The results of some of Jensen's measurements are set out in table 3, from which it can be seen that the compounds fall into two groups. In the first, where A is represented by various symmetrically substituted arsines, phosphines, and stibines, and X by NO2, C1, Br, etc., the moments are either approximately zero or around 8-12 Debye units. The compounds with zero moment are 9
I am indebted t o Mr. A. Maccoll for the compilation of this table.
155
STEREOCHEMISTRY O F SQUARE COMPLEXES
obviously the trans-forms and those with the large moment, the cis-forms. In PtClZ((CHs)ds)z, for example, the resultant of the three As-C bond moments is directed along the Pt-As bond irrespective of whether there is, or is not, TABLE 3 Dipole moments of platinous compounds of the type PtA& Jensen (71, 73) CObCPOUND
'
COMPOUND
DIPOLE MOMENT
Debye units
Debye units
trans- [PtBrz(Et 3P)21. . . . . . . . . . .
trans-[PtI~(Et3P)~]. .... trans-[PtC12(Pr3P)2].... trans- [PtClz (Et AS) 21. ......... trans-[PtI~(EtsSb)~] ........... t ~ ~ n ~ - [ P t C l z ( B u 3 P.).z.].. . . . . .
trans-[PtC12(Pr&3)2]. . . . . . . . . . . trans-[PtClt ( B U Z S ) ~ ] trans-[Pt (NO?)*(Pr2 truns-[PtBrt (Et2S)zI 1
Et
=
0
c i s - [ P t B r ~ ( E t s P ).~.]... . . . . . . . . cis-[PtIz(EtaP)z].
0 0 0
........... cis- [PtC12 (Et 3Sb)2 1 . . . . . . . . . . . . ci~-[PtCIz(BuaP)z] .............
2.35 2.35 2.48 2.26
DIPOLE MOMENT
'I
1 ~
~
cis-[PtCl*(Pr~S)~] ............. cis-[PtC12(Bu~S)~]. ... cis-[Pt(N0~)2(Pr2S)~]. ......... cis-[PtBr*(EtnS)~]. ...
11.2 8.2 11.5 10.5 9.2 11.5 9.5 13.1
ethyl; Pr = propyl; Bu = butyl.
FIG.3. Trans-planar structure for [PtC12((CH3)3As)2]
rotation about the Pt-As bond. The only reasonable structure for [PtC12((CH3)3As)2]which could give zero moment is the trans-planar (figure 3). I n the second group, which comprises the thioether complexes, there is a large difference between the CY- and P-forms, the dipole of neither form being zero. However, it is concluded that the compounds with the larger moments are the cis-forms and those with the smaller the trans-forms. If we make the
156
DAVID P. MELLOR
reasonable assumption that in trans-[Pt( (C2H6)zS)2C12] the platinum bonds are coplanar, the small moment of the molecule can be accounted for in terms of the known configuration of tercovalent sulfur. Indeed, the moment of the transform may be looked upon as supplying further proof that the three sulfur valence bonds are not coplanar. Figure 4 will make it clear that the S-C moments are not directed along the direction of the Pt-S bond, and that the molecule as a whole must have a resultant dipole moment,-in other words, the molecule in figure 4 has a trans-cis structure. Jensen's work provides a neat physical method for determining the configuration of metal complexes and a t the same time confirms the results of chemical methods of attack on this problem. More important still, it shows that the complexes are strictly planar, and in fact constitutes one of the strongest pieces of evidence against any suggestion of a pyramidal configuration. From the relevant dipole it is estimated that the platinum atom cannot, a t most, be more than 0.08 A. from the plane of the four attached groups. Dipole measure-
tats
FIG.4 ments do not tell us what the magnitudes of the C1-Pt-As bond angles are, and hence do not enable us to say whether the coordination is square
or not. Departures from angles of 90' do not disturb the centrosymmetry of the complex.
VII. THE CRYSTALSTRUCTURE OF PLATINOUS COMPOUNDS When based on sufficient reliable intensity data and carried through to completion, x-ray crystal analysis may be looked upon as a final court of appeal in structural problems. Often, however, a crystal may be so complicated and the difficulties associated with its determination so great that only a partial solution can be obtained. Evidence from crystal-structure analyses must therefore be given varying weight according to the completeness of the analysis, always
157
STEREOCHEXISTRY O F SQUARE COMPLEXES
assuming of course that each stage has been reliably determined. Several stages in an analysis may be recognized: (1) The determination of the size of the unit cell and the number of molecules contained therein. As a rule the number of molecules per unit cell determines to a large extent the complexity of a structure. If there happen to be a large number of molecules, in themselves fairly complicated, the task may prove extremely difficult, if not impossible, a t this stage of the development of x-ray technique. An investigation terminated a t the first stage is of little use, although on several occasions attempts have been made to use the fact that one of the dimensions of a unit cell is very small, to prove the existence of some planar structure. Conclusions based on such slender evidence must be accepted with caution, because a small cell dimension may be explained in some other way." ( 2 ) The second stage leads to the determination of the space group which describes fully the symmetry of the crystal structure. Once the number of molecules per unit cell and the space group have been determined, it is possible to state the symmetry elements of the molecule itself. By these means the following molecules (or ions) have been shown to have a center of symmetry: (a) bis(salicyla1doxime)platinum (25); ( b ) bis(dimethylsu1fine)dichloroplatinum (27) ; (c) bis(ethy1enediamine)platinous ion (26). In d l these instances centrosymmetry is consistent with trans-planar structures for the complexes, but not with tetrahedral structures. (3) The third and final step involves finding the exact location of each atom in the unit cell, and as a rule this becomes increasingly difficult as the number of parameters required to fix these positions increases. Relatively few complete structure determinations have been made on platinous compounds and they have, for the most part, been confined to fairly simple structures. Nevertheless the results are of great stereochemical interest. The structures of the following substances have been completely determined and in every instance square complexes have been found : (1) KJPtC141
(figure 5)
(2) [Pt(KH3)4]C12H20(figure 6)
(3) KZ[P~(C~OZSZ)ZI (figure 7) (4) PtC3zHi~Xs
(5) PtS
(figure 8)
The first substance examined mas KZ[PtC14] (38), which along with KZPdC14 belongs to the tetragonal system and has one molecule per unit cell. Because K2[PtC14] and [Pt(NH3)4]C12form the starting materials for the preparation of many of the isomeric ammines discussed earlier and in this sense are key sub10 Because one of its cell dimensions is small as compared with others, i t has been suggested (31) that the compound Co(py),Clz contains square-cobrdinated Co" Magnetic evidence is against this suggestion, and the small cell dimension can be explained on the basis of an octahedral structure (101).
.
158
DAVID P. M!3LLOR
stances, their structure will be discussed in detail. KzPtClr possesses an unusually simple strncture, involving as it does but one parameter, a parameter to fix the position of the chlorine atom in the plane of the platinum atoms, as shown in figure 5. The positions of the potassium and platinum atoms are fixed uniquely by symmetry. In the absence of any crystal faces indicating the contrary, Dickinson assumed that the crystal belonged to the holohedral class and his final structure determination in a sense justified this assumption. Nevertheless it is always useful to have-some independent evidence of the absence of lower symmetry in the crystal. Such has been provided by piezoelectric tests (142) and pyroelectric tests (106), both of which show that the crystal does not lack a center of symmetry.
FIQ.5. The structure of K2[PtClrl
The isomorphous palladium compound has been reinvestigated with more recently developed x-ray technique by Theilacker (142), and Dickinson's (38) earlier work has been completely confirmed. From his intensity data Theilacker concludes that the palladium atom cannot, a t most, he more than 0.2 A. out of the plane of the four chlorine atoms, and that since this is not much greater than the experimental error it is practically certain that all five atoms are strictly coplanar. If one accepts this evidence for the planar structure of the complex, then it fo~~ows immediately from the tetragonal symmetry of the crystal that the four Pt-CI bonds are a t exactly go", and that the PtC1,-- complex is definitely square. The stereochemical implications of this crystal structure determination were not a t first fully realized. They are, however, most important, because for the first time accurate information of the orientation of
STEREOCHEMISTRY O F SQUARE COMPLESES
159
the four platinum bonds became available. One or two additional points about the structure are worth noting. Firstly, the strong negative double refraction of the crystal (106) is consistent with the structure attributed to it. Several small discrepancies between observed and calculated intensities in Dickinson’s work on K2PtC14can be explained on the basis of the anisotropic thermal motions of the platinum atoms (66). However, the thermal oscillations of platinum atoms are about fixed mean positions so that no modification of the square structure is required.
ka
O !
4
J,
FIG. 6. The structure of [ P d ( N H d 4 l C l ~ H ~and 0 of [Pt(SHs)4]C12H20 (Strukturberzcht)
The second relatively simple compound, [Pt(NH3)4]C12H20,was first investigated by Cox (24), mho assigned to it a structure the same in essentials as that of K*PtCl4,--tvith the two chloride ions occupying the positions of the two potassium ions and the [Pt(NH3)4]++that of [PtCL]--. Subsequent work has shown that while the essential features of the [Pt(NH3)4]++complex are as Cox reported them (the four platinum valences coplanar and directed towards the corners of a square), the structure of the crystal is a little more complicated, requiring for its proper description a two-molecule unit cell (37) . I 1 The two-molecule cell arises from the small rotations of the [Pt(KH3)4]++complexes and chlorine atoms about the c-axis (figure 6). Robertson and Woodward’s (136) analysis of platinous phthalocyanine is one of the most extraordinary x-ray analyses carried out to date, but it adds little to the general theme developed here. The phthalo11 Dickinson’s most detailed work actually related to the isomorphous palladous compound Pd (NH?)cCln.H20.
160
DAVID P. MELLOR
cyanine molecule is a huge planar molecule and any metal atom attaching itself t o i t by the four pyrrole nitrogen atoms must do so by coplanar bonds. In the crystals so far mentioned, the complexes have all belonged to the k i t e class. In the last crystal of the group, viz., PtS (16), we find a new type,an infinite complex extending throughout the crystal in three dimensions. Platinum and sulfur maintain a coordination number four by appropriate sharing of atoms; a portion of an infinite chain structure can be seen in the unit cell of PtS shown in figure 8.
0 0
FIG.8. The structure of PtS
Fmally we come to several structures which have not been directly determined but which may be inferred from the complete crystal analysis of an isomorphous crystal. Square-coordinated metal complexes are found in each, and it must suffice merely to list them with the completely analyzed crystal indicated in brackets:
(a) Ba[Pt(CN)d4H20
((Bami(CN)4]4HzO))
(b) NadPt(CN)d3H20
{ (Na2[Ni(CN)d3HzO)) (15)
(c) PtClz
{ PdCL )
(15)
(d) PtO
WOl
(113)
(13)
161
STEREOCHEMISTRY O F SQUARE COMPLEXES
SUBSTANCE
DOUBLE REFRACTION
SYKMETRP
KI[PtC14].. . . . . . . . . . . . . . . . .Tetragonal Ba[Pt (CIi)414H20. . . . . . . . . . -Monoclinic Rlg[Pt ( C S ) 417H20. . . . . . . . . .Tetragonal
1.683 1.6706 1.561 1.6237 LiK[Pt(CN)4]3HIO... . . . . . . Orthorhombic trans-[Pt(py)NH3CL].. . . . . . \Triclinic 1.653 cis-[Pt ( P Y ) N H ~ C ~. .I .].. . . . . ? 1.624 K[PtC13C2H4]Hn0. . . . . . . . . . Monoclinic -0.137 1.732 >1.79 >-0.166 1.717 >-0.090 1.702 >?0.079 High and negative 1.613 1.637 -0.090 Extremely high 1.779 (1.80) -0.269 1.732 1.79 -0.135 1.696 1.750 -0.126 1.677
BEFEBENCES
(106)
(153) (55) (153) (56) (56) (78) (78) (15) (12) (32) (155) (155) (155)
The striking double refraction of crystals of the isomorphous series CaCOs, NaN03,and ScB03,for which the presence of parallel planar X03ions is responsible (lo),suggested that similar optical properties would be found among crystals containing square-coordinated Pt”. According to the results of x-ray analysis the square [PtCl&- groups in K2PtC14 are all arranged parallel to 001. If this arrangement is correct, the crystal should show strong negative double refraction, which indeed it does (106). Had the four chlorine atoms been tetrahedrally arranged about the platinum, one would have anticipated a small double refraction of the same order as that found in crystals containing SOT-, PO,--, C104, etc. Information about the crystal optics of other platinous compounds is summarized in table 4,from which it can be seen that the double refraction is high
162
DAVID P. MELLOR
throughout. The high double refraction of Zeise’s salt, K[PtC2H4CI3],is interesting; it is not known just how the ethylene is incorporated in the complex, but the complex undoubtedly has a square configuration. The strength of the double refraction of a crystal will depend on mutual arrangement of any anisotropic units it may contain, being greatest, and negative, when they are all parallel to one another as in CaC08and K2PtC14. Certain arrangements of planar groups result in positive double refraction, as in bastnaesite (154) ; with other arrangements, less probable no doubt, it is conceivable that a very low double refraction could be produced. Thus, while high double refraction undoubtedly indicates the presence of highly anisotropic units in a structure, low double refraction does not necessarily mean that such units are absent. In this connection another situation which may arise must be kept in mind. Double refraction is subject to dispersion,-that is, it varies with wave length,-and one may just happen to choose, for making a measurement, a wave length where the double refraction is low or at a minimum. This point is well brought out in Brasseur and Rassenfosse’s recent (12,15) extensive studies of the crystal optics of a whole series of complex cyanides of the types Ba[Me (CN) 4]4Hz0, Ca[Me (CN)4]5Hz0,Sr[Me(CN)4]4H20,and Naz[Me (CN)413Hz0, where Me = Pt, Pd, and Ni. Without exception, these substances show high double refraction, and for all except three, Ba[Pt(CN)4]4H20, Mg[Pt(CN)& ?”&, and Ca[Pt(CN)4]5H20,the sign of the double refraction is negative. A complete crystal-structure analysis of Ba[Ni(CN)4]4H20 reveals a structure which accords with high negative double refraction. The positive sign of the isomorphous platinum compound is an extremely puzzling anomaly for which no explanation has yet been given. There is little doubt about the observations on the positive sign, since the same results have been reported by several workers. It would seem that Bragg’s theory of the origin of double refraction of planar complexes needs further refinement if it is to take account of these platinum compounds. MAGNETIC ANISOTROPY
Practically nothing has been done on the diamagnetic properties of platinous compounds, but it can reasonably be expected that, like CaC03, NaN03, etc., they will show pronounced anisotropy. Some very early observations on Ca[Pt(CN)4]5H20were made by Grailich (58), who reported that the direction of greatest diamagnetic susceptibility is parallel to the c-axis of the (orthorhombic) crystal. This would place the plane of the [Pt(CN)4]-- group approximately perpendicular to the c-axis, whereas the optical properties suggest a different orientation. The crystal optics of the isomorphous nickel compound, Ba[Ni(CN)4]4HzO,place the plane of the [Xi(CK)&- group approximately parallel to the c-axis, in qualitative agreement with the observations on the diamagnetic anisotropy of the platinum compound. Further work, possibly along the lines of that of Born (9) and Hylleraas (68) on quartz and calomel, is needed to clear up the anomalous behavior of these platinum compounds.
STEREOCHEMISTRY O F SQUARE COMPLEXES
163
IX. DIRECTIVE INFLUENCES IN THE REACTIONS OF SQUARE COMPLEXES A.
Trans
ELIMINATION
In the light of the crystal-structure determinations of KZ[PtC14] and [Pt(NH3)4]ClzHz0it is of interest to note at this stage certain features of the reactions involved in the formation of the isomeric diammines. The most important of these is the process of trans elimination discovered by Werner. So far, little has been done towards providing a satisfactory understanding of this phenomenon, and all that will be attempted here will be to formulate some of the problems that arise. At the outset it is obvious that trans elimination cannot be a perfectly general reaction, because although it provides an explanation of some of the transformations, cis elimination must be invoked to explain others. ci
f
4+CI FIG. 9
Let us consider first the reaction responsible for the discovery of trans elimination : 2HC1
[Pt(NH3)4]++ ---+ [Pt(NHs)zC12]
trans If we regard the elimination of the two molecules of ammonia as occurring simultaneously, we might suppose that as the two trans molecules depart, two chlorine atoms enter the trans octahedral positions to form a new trans square complex, as shown in figure 9. But matters are not so simple as this. As already pointed out, the two reactions below proceed stepwise : [Pt(NH3)4]++
c% [Pt(NH3)3Cl]+ -% [Pt(NH3)2Cla] P
”
[PtCI4]-- -L
[PtSH3C13]- NHL[Pt(NH3)2Clz] a
164
DAVID P. MELLOR
If the complexes were tetra__edralit would be impossible to account the two different end products of these reactions. Let us imagine we have a square complex
into which another X is to be substituted for one of the A's. In what circumstances does the first X group direct the second one entering; to the cis position as in [PtNH3C13]- or to the trans position as in [Pt(NH3)aCl]+? At first sight it might seem as though X groups are trans directing when present in cationic complexes. Pinkard, Saenger, and Wardlaw (127) have studied very thoroughly the elimination reactions occurring with tetrammines containing ammonia, pyridine, and hydroxylamine, and in every instance the reaction indicates trans elimination.
YH3\
1 i
Pt/py
PY/
\ NHz OH
LCl/ NH3
NHzTt/"3
'
NHz OH
\PY
J
e1
\ / ' / \
c1
+ 2NHzOH
3"
e1 NHZ OH
NHZ OH
I
+ NHzOH +
3"
It will be seen that no other X group than C1 was investigated, and as far as the author is aware no systematic work has been done along these lines. However, the behavior of certain nitro palladium compounds is worthy of note. From the evidence available (91) it would seem that the NO2 group is directed to the cis position when it enters the complex [Pd(NH3)J402]+ and to the trans position when i t enters the complex [Pd(NH3)(NOz)3]-,which is just the opposite of the behavior of chlorine in the above platinum complexes. Two examples will suffice to show that the charge on the complex as a whole
STEREOCHEMISTRY O F SQUARE COMPLEXES
165
is not the factor determining directive influences. When [Pt(NH3)2(dipy)]Cl~ is treated with hydrochloric acid, the cis ammonia molecules are eliminated. This is perhaps hardly a fair test case, because if ammonia molecules are to be eliminated there is no choice but cis elimination. A more convincing case is the one discovered by Jensen (71) who, in the course of his dipole-moment investigations, found that when an aqueous solution of KzPtC14 is treated with four molecules of triethylphosphine, a colorless solution of [Pt { P(CzH5)3}4]C1~is ( p = 10.7 D) formed. This solution on standing deposits cis-[Pt{P(C~H5)3]~C12] and the only way this latter substance can be formed is by cis elimination from [Pt{P(Cd%)3)$ 2 1 2 . B.
cis
ELIMINATION
The reaction between ammonia and the [PtC14]-- ion seems typical of many amines. Cis elimination from this ion is known to occur with ethylamine, pyridine, hydroxylamine, aniline, etc., but again the reaction is not a perfectly general one. Some very interesting work in this field has been published by Tscherniaev and his school (57, 145). One of their most important findings is that order of substitution plays an important part in some complexes. Thus it was found that when ethylene is passed through Cossa’s potassium salt, K[Pt(NH3)CL], cis-[Pt (NH3)(CzH4)C12]is formed. On reversing the order of introduction of the groups, i.e., by treating Zeise’s salt, K[Pt(C2H4)Cl,], with ammonia, trans-[Pt(NH3)(CzH4)ClZ] is formed. Similar behavior was observed on substituting carbon monoxide for ethylene, but the effects of ordered substitution are confined to unsaturated substances like ethylene and carbon monoxide. It does not, for example, make any difference whether K[Pt(py)Cla] is treated with ammonia or K[Pt(NH3)C13] with pyridine, cis-[Pt(py) (NH,)C12] is the result. Another interesting but less common example of a directive influence is to be found in the complex ions formed with certain optically active diamines like Z-cyclopentane-trans-1,2-diamine(l-cptn). Jaeger and ter Berg (69) have shown that any attempt to introduce a second molecule of the dextro base into [Pt(Z-cptn)Clzlfails ; the only compounds obtained were ~ - [ P t ( l - c p t n ) ~ ] C and l~ L - [ P(~d - ~ p t n ) ~ ] where, C l ~ curiously enough, the sign of the rotation of the complex ions is opposite to that of the constituent optically active bases. These authors found no evidence for the existence of the meso form [Pt(d-cptn)(Z-cptn)]Clz. I n order to describe the curious propeller-like structure of these optically active complex ions, Jaeger and ter Berg have proposed to call them “pterotactic” complexes. The preceding examples are sufficient to show that the directive influences in substitution in square complexes present an interesting problem for the theoretical chemist.
X. THEUNIVERSALITY OF
SQUARESTRUCTURE AMONG PLATINOUS COMPOUNDS It now remains to consider whether the square structure is universal among Pt” compounds and as characteristic of that atom as the tetrahedral structure is THE
166
DAVID P. MELLOR
of carbon. The cases put to the test in physical investigation are necessarily few in number. As there is now no doubt about the origin of the geometrical isomerism, a better idea of the extent of the occurrence of the square structure can be gained from a brief survey of isomeric forms. Justification for extrapolating to cover all platinous compounds will be found in the quantum theory of the directed valence bond. VARIETIES O F ISOMERIC SQUARE COMPLEXES
It is a simple matter to draw up a scheme showing the types of isomeric complexes which are possible on the assumption that Pt" is square coordinated, and it is interesting to see how far these possibilities have been realized in practice. At the same time the scheme mill give some idea of the complexity introduced into the chemistry of Pt" by its habit of forming square bonds. All finite mononuclear complexes of quadricovalent Pt" must fall into one or other of these classes: (a) [PtA4]++, ( b ) [PtA3X]+, ( c ) [PtA2X2I0, ( d ) [PtAXJ, ( e ) [PtX4]--. I n the scheme adopted the following symbols have been used: ( 1 ) A, B, C, etc., to represent neutral molecules, e.g., KH3, C5H6N, KH20H, N2H4, CH3NH2, CzHJSHz, AsC13, P(CH3)3,(C2H&S, etc. (2) AB to represent an unsymmetrical bidentate chelate group attached by two coordinate links, e.g., isobutylenediamine. ( 3 ) AZ to represent an unsymmetrical chelate group attached by a coordination link and one primary link, e.g., glycine. (4) X , Y, Z, . . . etc., to represent a negatively charged atom or group such as C1-, CN-, NO,-, OH-, etc.
[;
Class A . Non-electrolyte or uncharged complexes
[:
Pt
.
Pt
;I
[;
Pt
:1'
,
AI,2
[;
:[
Pt
:I" [; ;I [: Pt
Pt
A5,B
;4, 4
Pt
